#asahi-re · 2022-10-08 14:24
Jamie[m]they're all mapped directly, the only thing i think we would need from firmware (except as an optimisation) would be for the interrupts to be forwarded#asahi-re · 2022-10-08 14:23
Jamie[m](or realistically potentially interfacing directly with the registers without going through the firmware)#asahi-re · 2022-10-08 14:22
Jamie[m]given the messiness of the firmware interface#asahi-re · 2022-10-08 14:22
Jamie[m]potentially#asahi-re · 2022-10-08 09:05
Jamie[m](running the fw under single-stepping, that is, including in interrupts, and still decoding video successfully)#asahi-re · 2022-10-08 09:04
Jamie[m]hehehe i think it's working#asahi-re · 2022-10-08 07:09
Jamie[m]i wonder how slow that is, and how much performance wiggle room we have before hitting timeouts#asahi-re · 2022-10-08 07:08
Jamie[m]so in theory, we could have it run the whole firmware single-stepped, inspecting opcode+registers to trace relevant actions#asahi-re · 2022-10-08 07:07
Jamie[m]which is that the debugmonitor can perform single-stepping#asahi-re · 2022-10-08 07:07
Jamie[m]not using the dwt at all#asahi-re · 2022-10-08 07:07
Jamie[m]there might be an alternative though#asahi-re · 2022-10-08 07:06
Jamie[m]and we've lost all information#asahi-re · 2022-10-08 07:06
Jamie[m]but as it is, i think a branch could happen in the middle#asahi-re · 2022-10-08 07:05
Jamie[m]if it was one cycle quicker, we could look back an instruction and parse the opcode, then inspect the relevant register state to figure out what the peripheral access address+data was#asahi-re · 2022-10-08 07:04
Jamie[m]if i set a watchpoint to happen on a write, the write happens, the instruction AFTER IT happens, and then my debugmonitor is hit#asahi-re · 2022-10-08 07:03
Jamie[m]damn, this isn't going to work in a straightforward way#asahi-re · 2022-10-07 15:07
Jamie[m](it's slow because i'm hooking the firmware writes, so it takes like 5-10 seconds to write the firmware)#asahi-re · 2022-10-07 15:07
Jamie[m]must be running up against a timeout of some sort then#asahi-re · 2022-10-07 15:06
Jamie[m]but then it started decoding video properly!#asahi-re · 2022-10-07 15:06
Jamie[m]oo i was just about to say "but it isn't decoding video properly"#asahi-re · 2022-10-07 15:06
Jamie[m]ok it's definitely booting the custom firmware and running the pre-init code#asahi-re · 2022-10-07 15:06
Jamie[m](i changed nothing)#asahi-re · 2022-10-07 15:06
Jamie[m]ayy got it unbroken somehow#asahi-re · 2022-10-07 15:02
Jamie[m]noo host reboot didnt fix it#asahi-re · 2022-10-07 14:44
Jamie[m]and decoded a video#asahi-re · 2022-10-07 14:44
Jamie[m]anyway right before this issue i'm 99% sure i successfully booted my patched firmware#asahi-re · 2022-10-07 14:44
Jamie[m]real solution here is for the shop to ship my AM5 motherboard#asahi-re · 2022-10-07 14:43
Jamie[m]yeah i think the host is just cursed#asahi-re · 2022-10-07 14:43
Jamie[m]which hopefully rules out cable issues#asahi-re · 2022-10-07 14:42
Jamie[m]the cable i just switched to is thunderbolt rated#asahi-re · 2022-10-07 14:41
Jamie[m]what could i have possibly changed to break this#asahi-re · 2022-10-07 14:41
Jamie[m]ugh i don't now how to continue working lol#asahi-re · 2022-10-07 14:39
Jamie[m]nooo it happened again#asahi-re · 2022-10-07 14:39
Jamie[m]oh good now i'm getting errors from dmesg as well#asahi-re · 2022-10-07 14:39
Jamie[m]it's not normally this bad i swear#asahi-re · 2022-10-07 14:36
Jamie[m]and the room got more cold#asahi-re · 2022-10-07 14:36
Jamie[m]except maybe the mac battery got more full#asahi-re · 2022-10-07 14:36
Jamie[m]i haven't changed anything#asahi-re · 2022-10-07 14:36
Jamie[m]man #asahi-re · 2022-10-07 14:35
Jamie[m]would that be good or bad? don't have one in any case since this mac came with magsafe#asahi-re · 2022-10-07 14:35
Jamie[m]nah#asahi-re · 2022-10-07 14:34
Jamie[m]i think there's something cursed about usb signal integrity on my desktop#asahi-re · 2022-10-07 14:33
Jamie[m]proxy keeps dying with repeated usb-dwc3@b02280000: Host cleared EP 0x82 stall :(#asahi-re · 2022-10-07 14:33
Jamie[m]noo my usb has stopped working#asahi-re · 2022-10-07 14:20
Jamie[m]oh it took like 10 tries to do it it but it worked eventually#asahi-re · 2022-10-07 14:20
Jamie[m]ah, i think there might be a timeout on firmware loading#asahi-re · 2022-10-07 13:49
Jamie[m](i.e. a generic python module/class-reloading feature)#asahi-re · 2022-10-07 13:49
Jamie[m]might be generic python knowledge i'm lacking#asahi-re · 2022-10-07 13:49
Jamie[m]but idk what the entrypoint to trigger it is#asahi-re · 2022-10-07 13:49
Jamie[m]i'm pretty sure there's tracer-reloading functionality though, i can see all sorts of plumbing to make it work